24小时热门版块排行榜    

查看: 7580  |  回复: 72
当前主题已经存档。

灯塔守望者

铁杆木虫 (正式写手)

[交流] 在线讨论:使用vasp,exciting计算材料的磁性问题

受csfn兄的号召,引用wuchenwf兄弟的话“我个人水平有限,所以在此抛砖引玉。”
本人主要关注稀磁半导体的磁性问题,常用的软件主要有vasp, exiting(FLAPW), abinit, pwscf也会使用一些。希望在此和各位朋友共同探讨遇到的问题。
在此先介绍几个优秀的开源软件,大家熟悉的abinit, pwscf就不说了。

1 exciting (http://exciting.sourceforge.net/)Features
General
• FP-LAPW basis with local-orbitals
• APW radial derivative matching to arbitrary orders at muffin-tin surface (super-LAPW, etc.)
• Arbitrary number of local-orbitals allowed (all core states can be made valence for example)
• Every element in the periodic table available
• Total energies resolved into components
• LSDA and GGA functionals available
• Core states treated with the radial Dirac equation
• Simple to use: just one input file required with all input parameters optional
• Multiple tasks can be run consecutively

Structure and symmetry
• Determination of lattice and crystal symmetry groups from input lattice and atomic coordinates
• Determination of atomic coordinates from space group data (with the Spacegroup utility)
• XCrysDen and V_Sim file output
• Automatic reduction from conventional to primitive unit cell
• Automatic determination of muffin-tin radii
• Full symmetrisation of density and magnetisation and their conjugate fields
• Automatic determination and reduction of the k-point set

Magnetism
• Spin polarised calculations performed in the most general way: only (n(r); m(r)) and (vs(r); Bs(r)) are referred to in the code
• Spin symmetry broken by infinitesimal external fields
• Spin-orbit coupling (SOC) included in second-variational scheme
• Non-collinear magnetism (NCM) with arbitrary on-site magnetic fields
• Fixed spin-moment calculations (with SOC and NCM)
• Spin-spirals for any q-vector (conical spirals are still experimental)

Plotting
• Band structure plotting with angular momentum character
• Total and partial density of states with irreducible representation projection
• Charge density plotting (1/2/3D)
• Plotting of exchange-correlation and Coulomb potentials (1/2/3D)
• Electron localisation function (ELF) plotting (1/2/3D)
• Fermi surface plotting (3D)
• Magnetisation plots (2/3D)
• Plotting of exchange-correlation magnetic field, Bxc (2/3D)
• Plotting of ∇⋅Bxc (1/2/3D)
• Wavefunction plotting (1/2/3D)
• Electric field (E=-&nabla V) plotting (1/2/3D)
• Simple scanning tunnelling microscopy (STM) imaging based on the local density of states (LDOS) (experimental)

Forces and phonons
• Forces - including incomplete basis set (IBS) and core corrections
• Forces work with spin-orbit coupling, non-collinear magnetism and LDA+U
• Structural optimisation
• Phonons for arbitrary q-vectors (experimental)
• Phonon dispersion and density of states
• Thermodynamic quantities calculated from the phonon DOS: free energy, entropy, heat capacity
• Phonon calculations can be distributed across networked computers
• Electron-phonon coupling matrices
• Phonon linewidths
• Eliashberg function, α2F(ω)
• Electron-phonon coupling constant, λ
• McMillan-Allen-Dynes critical temperature, Tc

Advanced
• Exact exchange (EXX) optimised effective potential (OEP) method (with SOC and NCM) (experimental)
• EXX energies (with SOC and NCM) (experimental)
• Hartree-Fock for solids (with SOC and NCM) (experimental)
• LDA+U: fully localised limit (FLL), around mean field (AFM) and interpolation between the two; works with SOC, NCM and spin-spirals (experimental)
• Reduced density matrix functional theory (RDMFT) for solids (experimental)

Miscellaneous
• Mössbauer hyperfine parameters: isomer shift, EFG and hyperfine contact fields (experimental)
• First-order optical response
• Kerr angle and Magneto-Optic Kerr Effect (MOKE) output (experimental)
• Generalised DFT correction of L. Fritsche and Y. M. Gu, Phys. Rev. B 48, 4250 (1993) (experimental)
• Energy loss near edge structure (ELNES)
• L, S, and J expectation values
• Effective mass tensor for any state
• Equation of state fitting (with the EOS utility)

Programming
• Clean, simple code structure - ideal for development
• OpenMP parallelisation over k-vectors
• Strict Fortran 90 compliance
• Only one language used
• Free-form style input file
• Full LaTeX documentation included with every subroutine

2 OpenMX (http://www.openmx-square.org/whatisopenmx.html)
OpenMX (Open source package for Material eXplorer) is a program package for nano-scale material simulations based on density functional theories (DFT) [1], norm-conserving pseudopotentials [2,20,21], and pseudo-atomic localized basis functions [23]. Since the code is designed for the realization of large-scale ab initio calculations on parallel computers, it is anticipated that OpenMX can be a useful and powerful tool for nano-scale material sciences in a wide variety of systems such as biomaterials, carbon nanotubes, magnetic materials, and nanoscale conductors. The distribution of the program package and the source codes follow the practice of the GNU General Public License (GPL) [39], and they are downloadable from http://www.openmx-square.org/


Features and capabilities of OpenMX Ver. 3.2 are as follows:

Total energy and forces by cluster, band, and O() methods
Local density approximation (LDA, LSDA) [2,3,4] and generalized gradient approximation (GGA) [5] to the exchange-correlation potential
Norm-conserving pseudopotentials [2,20,21]
Variationally optimized pseudo-atomic basis functions [23]
Fully and scalar relativistic treatment within pseudopotential scheme [10,19,13]
Non-collinear DFT [6,7,8,9]
Constraint DFT for non-collinear spin and orbital orientation [11]
Collinear LDA+U and non-collinear LDA+U methods [16]
Macroscopic polarization by Berry's phase [12]
Electric transport calculation by a non-equilibrium Green's function method
Divide-conquer (DC) method [28], generalized DC method, and Krylov subspace method for O() eigenvalue solver
Simple, RMM-DIIS [31], GR-Pulay [30], Kerker [32], and RMM-DIIS with Kerker's metric [31] charge mixing schemes
Exchange coupling parameter [14,15]
Optical conductivity
Charge doping
Uniform electric field
Full and constrained geometry optimization
NVE ensemble molecular dynamics
NVT ensemble molecular dynamics by a velocity scaling [17] and the Nose-Hoover methods [18]
Mulliken, Voronoi, and ESP fitting analysis of charge and spin densities
Analysis of wave functions and electron (spin) densities
Dispersion analysis by the band calculation
Density of states (DOS) and projected DOS
Flexible data format for the input
Completely dynamic memory allocation
Parallel execution by Message Passing Interface (MPI)
Useful user interface for developers
Evaluation of two-center integrals using Fourier transformation [27]
Evaluation of three-center integrals by a projector expansion method [24]
Solution of Poisson's equation using FFT [26]
Considerable functionalities are available for calculations of physical properties such as magnetic, dielectric, electric transport properties as listed above. Not only conventional diagonalization schemes are provided for clusters, molecules, slab, and solids, but also linear scaling methods are supported as the eigenvalue solver. Three calculation parts in OpenMX are mainly time-consuming:

Evaluation of Hamiltonian matrix elements
Solution of Poisson's equation
Diagonalization of the generalized secular equation
For the first and second parts, the computational time always scales as O() and O() for any eigenvalue solver, where  is the number of atoms, basis functions, or grid points. When the conventional diagonalization scheme (cluster and band methods) is used, the computational time for the third part scales as O(). On the other hand, the O() methods can solve the eigenvalue problem in O() operation in exchange for accuracy. For large scale calculations parallel execution by MPI is supported for parallel machines with distributed memories. Also all work arrays in the program codes are dynamically allocated with the minimum memory size required to an input file. The execution environment is unix and linux. For the execution of OpenMX, you are required to have pseudo-atomic basis orbitals and pseudopotentials. These input data can be calculated using ADPACK which is a program package for atomic density functional calculations. Conveniently, the data for several elements and ADPACK are available from a web site (http://www.openmx-square.org/). We are continuously working toward development. Motivated contributors who want to develop the open source codes are welcome. If so, the contact information is available in the above website.

[ Last edited by woshilsh on 2009-6-13 at 22:27 ]
回复此楼
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

csfn

荣誉版主 (知名作家)

优秀超版优秀版主优秀区长

兄弟,问3个问题
1,vasp在计算磁性的时候,oszicar中得到的磁矩和outcar中得到各原子磁矩之和不一致,在投稿的是否曾碰到有审稿人质疑,对于这个不一致你们一般是怎么解释的了?
2,另外,磁性计算应该比较负责。你应该还使用别的程序计算过磁性,与vasp结果比较是否一致,对磁性计算采用的程序有什么推荐。
ps:由于曾使用vasp和dmol算过非周期体系磁性,结构对磁性影响非常大,因此使用这两个程序计算的磁性要一致很麻烦。还不敢确定到底是哪个程序可能不可靠。
3,如果采用vasp计算磁性,对采用的方法和设置有什么推荐。
2楼2008-12-04 13:28:29
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

灯塔守望者

铁杆木虫 (正式写手)

★ ★ ★ ★ ★ ★
csfn(金币+6,VIP+0):many thanks
1,OSZICAR中得到的磁矩是OUTCAR中最后一步得到的总磁矩是相等的。总磁矩和各原子的磁矩(RMT球内的磁矩)之和之差就是间隙区的磁矩。因为有间隙区存在,不一致是正常的。
2,如果算磁性,全电子的结果更精确,我的一些计算结果显示磁性原子对在最近邻的位置时,PAW与FPLAW给出的能量差不一致,在长程时符合的很好。虽然并没有改变定性结论。感觉PAW似乎不能很好地描述较强耦合。我试图在找出原因,主要使用exciting和vasp做比较。计算磁性推荐使用FP-LAPW, FP-LMTO, FPLO很吸引人(不过是商业的),后者是O(N)算法。
3,使用vasp计算磁性,注意不同的初始磁矩是否收敛为同一个磁矩。倒没有特别要注意的地方,个人认为。

归根结底,需要一个优秀的交换关联形式出现。
3楼2008-12-04 14:04:13
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

gaofenglili

银虫 (小有名气)

您好!请问用VASP计算是否也是像计算DOS和能带一样要进行三步(结构优化,静态自洽计算,非自洽计算),然后看最后一步的出的磁矩呢?我用PAW势计算材料缺陷产生的磁性时,在最后一步给出WARNING: stress and forces are not correct ,请问会是什么原因呢?对结果影响大吗?

计算磁性必须考虑交换相关能吗?怎么实现呢?

多谢了!
4楼2008-12-04 14:25:45
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

灯塔守望者

铁杆木虫 (正式写手)

★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★ ★
jghe(金币+2,VIP+0):呵呵,支持!
csfn(金币+10,VIP+0):专帖主辛苦了,平安小红包慰劳:-)
你好,通常我都是在结构优化后得到总磁矩。
WARNING: stress and forces are not correct这个错误我很少遇到,没有深究过。你可以换个算法试试看。参见IBRION和ALGO命令. 如果你的结构离平衡结构很近,应该多结果影响小。
计算磁性必须考虑交换相关能吗?不明白你的意思。目前PBE,PW-91交换关联关联式描述磁性不够完美,需要更好的交换关联形式出现。
5楼2008-12-04 14:37:34
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

gaofenglili

银虫 (小有名气)

明白了,多谢您的帮助!
6楼2008-12-04 15:10:00
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

jghe

金虫 (著名写手)

有没有跟abinit 比较过,现在有很多人用abinit,我也想用,不知道算此行怎么样?
7楼2008-12-04 20:23:08
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

gaofenglili

银虫 (小有名气)

您好!我想问个关于半金属的问题:

  我计算后得到的DOS图, 在Ferimi 能级附近由明显的自选极化现象,其中一个图Fermi能级穿过minaority spin而不穿过majority spin,这是否能确定体系具有半金属性质?还有几个图Fermi能级明显穿过minaority spin但是也穿过majority spin的一点(不放大看不出来),这算不算半金属呢
8楼2008-12-05 10:31:37
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

灯塔守望者

铁杆木虫 (正式写手)

★ ★
wuchenwf(金币+2,VIP+0):xiexie
引用回帖:
Originally posted by jghe at 2008-12-4 20:23:
有没有跟abinit 比较过,现在有很多人用abinit,我也想用,不知道算此行怎么样?

abinit没有vasp计算准确的原因主要是赝势质量不如vasp的。我粗略比较过。如果两这都采用PAW,其磁矩很接近全电子,两者结果相差很小。如果是abinit+NCPP算磁性,给出的磁矩误差很大。abinit很优秀,功能很强大,现在需要提高PAW势文件的质量。相信一定和vasp媲美的。计算速度不如vasp,希望改进算法和支持更多的数学库函数。这是我个人的一点浅薄看法。
9楼2008-12-05 12:04:52
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

灯塔守望者

铁杆木虫 (正式写手)

★ ★
wuchenwf(金币+2,VIP+0):xiexie
半金属就是一个自旋态密度在fermi能上有状态数,另一个自旋态密度在fermi能上没有状态数。你说的那种体系算是半金属。采用标准DFT计算得到的half metal未必是真实的。
10楼2008-12-05 12:10:07
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖
相关版块跳转 我要订阅楼主 灯塔守望者 的主题更新
普通表情 高级回复 (可上传附件)
信息提示
请填处理意见